[Spectrum 2.0] Comments

[CresitelloDittmar, Mark]

Mireille,


On Wed, Jun 10, 2015 at 10:17 AM, Louys Mireille <mireille.louys at unistra.fr>
wrote:

>
>
>
>
>
>
>
>
> * Hi Mark, hi all, Many thanks for the update and corrections in the
> document which improves in clarity , namely for the observation , dataset
> Ids . Only minor suggestions below. My comments in slanted font:
> stdRefPosition: appears first in 3.10 and is defined in 6.4 in the
> datatypes. *
> * that would help to have a reference link to the definition on the first
> occurences in the text. *
>
> That is true for many elements of the model, I think that would be a good
addition to the newer documents
(DatasetMetadata and NDCube), which are earlier in the process.


> p67.
> *Dataproducttype enumeration list contains one literal equals to 'MIXED
> Some combination of the other options'.*
>
>
> *Are there any service implementations currently using this kind of data
> product type? The problem of *
>
> *combined , and or complex datasets is not supported by this Spectrum
> model. There will be a discussion in Sesto In DAL on the use of DataLink
> and existing models to tackle these situations . *
>

The 'MIXED' value was removed in the Dataset Metadata document.. where we
resolved the set shown here, and the set in ObsCore.


> p65.
> The concept of the ”nu L-nu” or ”lambda L-lambda” luminosity flux, or
> equivalently the
> luminosity per logarithmic energy interval L(log nu), is a distinct
> concept in the world of spectral
> energy distributions - and it’s a di erent concept from the bolometric
> luminosity, ff which has the
> same units. The UCD board has not yet approved a UCD expressing this
> concept; we have to
> use phys.luminosity and infer the concept from the units.
>
> *is this an implicit  request for a new UCD **? this can be taken into
> account in Semantics **, I sent a request for that.*
>

Yes, I suppose it is.. I don't know if a formal request was ever submitted
to the Semantics group.


> p97.
> • A Photometry Filter may not be serialized in the same file as the
> photometric sequence.
> There is no mechanism for serializing a reference to another extension
> within the same file.
> It may be referred to via the keyword holding the reference URI.
>
> *I am not sure I understand this paragraph : do we need an example here to
> show *
> *how to link an example of  a spectrum  (spectrumDM2.0 ) to a filter
> description from Filter in PhotDM?*
>
>
In the FITS serialization, one cannot serialize the filter in the same file
as the spectrum because
there is no FITS standard for referring to another extension of the same
file.  So, one must use the
keyword which maps to the URI pointer.  (PHID, I think).

The example serializations in the file are limited to the required set,
mainly because a complete serialization would be very large.. but yes, that
would be helpful.



>
> *RefPos vs referencePosition:*
> *all Frame in the text use the attribute name 'referencePosition' but the
> utype list uses 'RefPos' like in  CoordSys.FluxFrame.RefPos*
> *if the utype string does not matter why not homogeneise and use
> 'referencePosition'in the Utype string. *
> *This will be generated this way **in both cases : VODML serialisation or
> legacy utypes **, I suppose.*
>
>
The UType with 'RefPos' is consistent with the other frames, and earlier
'old-style' utype lists for the spectral model.  Its true, the migration to
a vodml representation will result in vo-dml tags that are more consistent
with the attributes.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ivoa.net/pipermail/dm/attachments/20150610/ad14c6b2/attachment-0001.html>